Add more testcases

[dota17]

Major changes:

• add test_family test case
• rewrite test_track_order test case. I think the origin test case are not very straightly to show the track_order feature, based in that test case we cant't discover the bug referred in track_order not working on root group #1577.

[codecov]

Codecov Report

Merging #1618 into master will increase coverage by 0.03%.
The diff coverage is n/a.

@@            Coverage Diff             @@
##           master    #1618      +/-   ##
==========================================
+ Coverage   88.47%   88.50%   +0.03%     
==========================================
  Files          17       17              
  Lines        2247     2254       +7     
==========================================
+ Hits         1988     1995       +7     
  Misses        259      259              

Impacted Files	Coverage Δ
h5py/_hl/base.py	95.63% <0.00%> (ø)
h5py/_hl/dataset.py	93.15% <0.00%> (+0.03%)	⬆️
h5py/_hl/group.py	96.80% <0.00%> (+0.05%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update a37eed8...ffba213. Read the comment docs.

[takluyver]

Thanks. The test for the family driver will be a useful addition.

It makes sense to test track_order on the root group of a file, but I don't think we need to throw away the existing track_order tests to add that; they're pretty well written as they are.

[takluyver]

Why did you scrap the .populate() method? The numeric vs alphabetical ordering is a neat way to test this (2 < 10 but '10' < '2'), and it was also checking that having a mixture of groups and datasets didn't interfere with the ordering.

[dota17]

uhh, the populate() method is well designed, but not good for test_group, the existing bug have proved this, it is not a very straightforward way to show track_order feature for group, we should focus on testing root group and sub group, but not mixing group and dataset, so I propose to write it like this:

def populate(self, g, track_order):
    for i in range(9,0,-1):
        if i % 2 == 0:
            self.f.create_group(str(i), track_order)  # create root group
        else:
            g.create_group(str(i), track_order)       # create sub group

@ut.expectedFailure
def test_track_order(self):
    grp = self.f.create_group('0', track_order=True)   # --->True
    self.populate(grp, True)  # --->True
    expected = ['8', '6', '4', '2', '0']
    # this will fail until issue 1577 fixed
    self.assertEqual(list(self.f.keys(), expected))
   
    expected = ['9', '7', '5', '3', '1']
    self.assertEqual(list(grp.keys(), expected))

def test_no_track_order(self):
    grp = self.f.create_group('0', track_order=False)
    self.populate(grp, False)
    expected = ['0', '2', '4', '6', '8']
    self.assertEqual(list(self.f.keys(), expected))
   
    expected = ['1', '3', '5', '7', '9']
    self.assertEqual(list(grp.keys(), expected))   

[dota17]

@takluyver Ready for your review.